New Omap3530 Clocked Higher?


gibberish

Active Member
Joined
Jan 26, 2008
Messages
865
Age
42
Location
Nottingham, UK
is it true that ti have discontinued the current processor? are the second+ batches of pandoras going to have a cpu clocked higher than the current ones?
is it an identical cpu, just clocked higher? or have they made some changes to allow to run at 700mhz?

in other words, are some first batchers going to be able to clock their pandora up to 900mhz or so, whereas some second batchers will be able to clock theres to 1ghz?

i apologise if this has already been asked/answered.
 
You can find some info here:
http://www.gp32x.de/board/index.php?/topic/51393-clockability/page__p__797767
 
Last edited by a moderator:
what i think is, since weve paid for the omap thats gonna be inside pandora, im sure the price of them has dropped, and before ordering the 4000 omaps 1-3 years later than customer paid, they shoulda bought ones that are worth what we paid....
 
DaMummy said:
what i think is, since weve paid for the omap thats gonna be inside pandora, im sure the price of them has dropped, and before ordering the 4000 omaps 1-3 years later than customer paid, they shoulda bought ones that are worth what we paid....
ahmm.... iirc, the soc's were bought and delivered ~ a year are ago. No chance to benefit for first-batchers.
 
Last edited by a moderator:
DaMummy said:
what i think is, since weve paid for the omap thats gonna be inside pandora, im sure the price of them has dropped, and before ordering the 4000 omaps 1-3 years later than customer paid, they shoulda bought ones that are worth what we paid....
Yes, just bin the 'old' ones. Boy you are a genius :blink:
 
Last edited by a moderator:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.
 
mcobit said:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.
No one knows what they found out. That's just one theory (and my personal favorite), and the extra cost of the 720 is in the extra test used to guarantee functionality at the higher clock speed, but the slower 600s should also probably clock to 720. This means the 600s coming off the line now should probably still be overclockable, but not guaranteed.
The other theory is that there is a difference, they're performing the 720mhz test on all chips, and the ones that pass go into the 720 bin and the ones that fail go into the 600, meaning the 600s coming off the line now really are just 600s, and will not reach 720mhz.
 
Last edited by a moderator:
I would just like to note that the iPhone 3Gs is downclocked and locked to run at 533 mhz. Can't change it even on a jailbroken phone. It's a read only thing Samsung/Apple did to a command that access the hz frequency. So, I would imagine the GPU to also be downclocked and locked at it's lower then rated frequency.

http://macthemes.net/forum/viewtopic.php?pid=519573

So, the Pandora's CPU is faster by default, and the GPU possibly around the same ;)
 
WizardStan said:
mcobit said:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.
No one knows what they found out. That's just one theory (and my personal favorite), and the extra cost of the 720 is in the extra test used to guarantee functionality at the higher clock speed, but the slower 600s should also probably clock to 720. This means the 600s coming off the line now should probably still be overclockable, but not guaranteed.
The other theory is that there is a difference, they're performing the 720mhz test on all chips, and the ones that pass go into the 720 bin and the ones that fail go into the 600, meaning the 600s coming off the line now really are just 600s, and will not reach 720mhz.

In addition to that theory, Was that the ones that could/may clock higher than 720, Could/may go into another bin (720+ bin),
Thus removing the possibility of people having the chance to clock above 720 to 900 or maybe more, Like we have seen on some of the dev
boards.

Trooper
 
Last edited by a moderator:
trooper said:
WizardStan said:
mcobit said:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.
No one knows what they found out. That's just one theory (and my personal favorite), and the extra cost of the 720 is in the extra test used to guarantee functionality at the higher clock speed, but the slower 600s should also probably clock to 720. This means the 600s coming off the line now should probably still be overclockable, but not guaranteed.
The other theory is that there is a difference, they're performing the 720mhz test on all chips, and the ones that pass go into the 720 bin and the ones that fail go into the 600, meaning the 600s coming off the line now really are just 600s, and will not reach 720mhz.

In addition to that theory, Was that the ones that could/may clock higher than 720, Could/may go into another bin (720+ bin),
Thus removing the possibility of people having the chance to clock above 720 to 900 or maybe more, Like we have seen on some of the dev
boards.

Trooper

Having a 720+ bin would only make sense if they prepare the release of a new version with a higher clock guaranteed and wanted to have those in stock. Then it would more be like 800+ or even 900+, they wouldn't bother with selling versions with 720 and 750Mhz or something like that at the same time.
I don't really see a market for 800Mhz+ omaps at the moment, as most smartphones aren't designed to handle the extra heat.
 
Last edited by a moderator:
mcobit said:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.

That's a false understanding, although a very easy one to make. They're still selling 600MHz parts, ie the ones that don't end in "72."
 
Last edited by a moderator:
Exophase said:
mcobit said:
No need to bin anything. As I understand it, they found out, that all chips can be clocked to 720 Mhz without problems, so they altered the specifications on the paper. We get the same processors as the 2nd batchers.

That's a false understanding, although a very easy one to make. They're still selling 600MHz parts, ie the ones that don't end in "72."

Bang on.

The problem will only arise with the second batch, but that's not to say that the first batchers will win on the MHz scores. Of the first batch, all will be clocked at 600MHz, and any number of the first batch processors will be able to over-clock to 720MHz - it's not guaranteed, they haven't been tested and rated. The second batch, assuming they buy the same CPUs, will probably be unable to achieve the 720MHz mark with any stability as they will have failed the tests at 720MHz.

D.
 
Last edited by a moderator:
Craig said he was signing the papers for 8000 more. Could that mean he put in the order for SoCs? It would be good to know if he decided to go with either 600mhz or 720mhz parts.
 
Dunny said:
Bang on.

The problem will only arise with the second batch, but that's not to say that the first batchers will win on the MHz scores. Of the first batch, all will be clocked at 600MHz, and any number of the first batch processors will be able to over-clock to 720MHz - it's not guaranteed, they haven't been tested and rated. The second batch, assuming they buy the same CPUs, will probably be unable to achieve the 720MHz mark with any stability as they will have failed the tests at 720MHz.

D.

The problem with that is going by that logic x86 CPUs should overclock very poorly since they're heavily binned. It's just not possible to fill the bins with losers when the demand for lower end CPUs is much higher than the percentage failing the higher bin tests. Since TI is only selling two speed grades it goes even further than this, since instead of having an extensive test to bin them to one of many speed grades they can just do a single pass/fail test. My guess is that TI will test some number of them based on demand for 720 units and the 600s will either be untested chips or 720s that failed. The percentage of 720s that fail is unknown, but it has to be low enough for TI to even bother selling these. In actuality, I don't think chips really vary that much and probably bell curve around some point (and in this case that point is probably higher than 720, especially given what we know about the chips out there), if they're all the same stepping and process anyway.
 
Last edited by a moderator:
WizardStan said:
The other theory is that there is a difference, they're performing the 720mhz test on all chips, and the ones that pass go into the 720 bin and the ones that fail go into the 600, meaning the 600s coming off the line now really are just 600s, and will not reach 720mhz.

Now that's a rather simplistic and negative view.

It's far more likely that they're testing power consumption at different speeds. This is basically a cellphone SoC, after all. All the 720mhz ones probably consume less juice, so they bump them up to 720mhz, and then they're both equal. ;)

Or maybe not.

P.S. My experience with 90nm AMD CPUs was the low power ones overclock worse. Not by much, but there was a few hundred mhz difference. Could be something similar going on here... or not. It's all speculation.
 
Last edited by a moderator:
We have placed the same order for the same chips. The 720 chips are *TWICE* the price. It is exactly the same chip it just *might* have been tested, but who knows what the test even is or how many they are testing? The lead time on the 720 is also about 6 months.

We will just stick to the 600mhz ones. It's not like they are not amazingly fast anyway. We will do some overclocking experiments on them though to see if there is any difference.
 
craigix said:
We will just stick to the 600mhz ones. It's not like they are not amazingly fast anyway. We will do some overclocking experiments on them though to see if there is any difference.

They were amazingly fast 16 months ago. They would still have been fast in Nov. 2008. They are no longer amazingly fast - there have been faster mobile devices announced now. The only thing that makes the Pandora stand above the upcoming competition is the control set.

D.
 
Last edited by a moderator:
Dunny said:
The only thing that makes the Pandora stand above the upcoming competition is the control set.
What about battery size, formfactor, openness, community, devs, etc.?
 
Last edited by a moderator:
Kramy said:
WizardStan said:
The other theory is that there is a difference, they're performing the 720mhz test on all chips, and the ones that pass go into the 720 bin and the ones that fail go into the 600, meaning the 600s coming off the line now really are just 600s, and will not reach 720mhz.

Now that's a rather simplistic and negative view.

It's far more likely that they're testing power consumption at different speeds.

We have rather simplistic data - i.e. almost none. We are told that they now have two products from one process instead of one, but the manufacturing process has not changed. So, if there have been no improvements, it seems it boils down to a very few select possibilities. The 720's are guaranteed to run at a higher clock rate than the 600's are guaranteed to run. Either the 600's fail some test to denote their lower speed (in which case 720's are more overclockable than First Batch SOCs) or they aren't tested, in which case we're likely talking about no detectable difference, on average.

mali said:
Dunny said:
The only thing that makes the Pandora stand above the upcoming competition is the control set.
What about battery size, formfactor, openness, community, devs, etc.?

Does the Pandora stand above, say, the Wiz in all those respects? I find it hard to judge other communities that I've never even seen, but it seems unlikely that the Pandora community is, and shall remain, THE BEST community bar none especially when we're so very mnuch a part of the GP32X community. IMHO.
 
Last edited by a moderator:
Back
Top